When Rodney Brooks talks about robotics and artificial intelligence, you should listen to him. Currently the Panasonic Professor of Robotics Emeritus at MIT, he also co-founded three major companies, including Rethink Robotics, iRobot, and his current endeavor, Robust.AI. Brooks also ran the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) for a decade, starting in 1997.
In fact, he likes making predictions about the future of AI and IT. Keeps a scorecard He told on his blog how well his work is going.
He knows what he’s talking about, and he thinks maybe it’s time to put a stop to the hue and cry about generative AI. Brooks thinks it’s impressive technology, but perhaps not as capable as many are suggesting. “I’m not saying LLMs aren’t important, but we have to be careful [with] “That’s how we evaluate them,” he told TechCrunch.
The problem with generative AI, he says, is that, while it’s perfectly capable of performing certain tasks, it can’t do everything a human can, and humans tend to overestimate its capabilities. “When a human sees an AI system performing a task, they immediately generalize it to similar things and overestimate the capability of the AI system; not just performance on it, but capability around it,” Brooks said. “And they’re usually way overly optimistic, and that’s because they’re using a model of a person’s performance on a task.”
The problem, he said, is that generative AI is not human or human-like, and trying to give it human capabilities is flawed. He says people think it is so capable that they even want to use it for applications that don’t make any sense.
Brooks offers his latest company Robust.ai, a warehouse robotics system, as an example of this. Someone recently suggested to him that it would be cool and efficient to tell his warehouse robots where to go by creating an LLM for his system. However, in his estimation, this is not a proper use case for generative AI and would actually slow things down. It is much easier to instead connect the robots to the stream of data coming from warehouse management software.
“When you have 10,000 orders and you have to ship them in two hours, you have to optimize for that. Language won’t help; it will only slow things down,” he said. “We have massive data processing and massive AI optimization technology and planning. And that’s how we fulfill orders faster.”
Another lesson Brooks has learned when it comes to robots and AI is that you can’t try to do too much. You need to solve a problem that is solvable and into which a robot can be easily integrated.
“We need automation in places that are already cleaned. So the example from my company is we’re doing really well in warehouses, and warehouses are really very confined. The lighting doesn’t change in those big buildings. There’s no stuff lying on the floor because people pushing carts could fall into it. There’s no floating plastic bags. And largely it’s not in the interest of the people working there to be malicious toward the robots,” he said.
Brooks explains that it’s also about robots and humans working together, so his company designed these robots for practical purposes related to warehouse operations, rather than creating a robot that looks like a human. In this case, it looks like a shopping cart with a handle.
“So the form factor we use is not like a walking human — even though I’ve built and distributed more human-like ones than anyone else. These look like shopping carts,” he said. “It has a handlebar, so if the robot has a problem, someone can grab the handlebar and do whatever they want with it,” he added.
After so many years, Brooks has learned that it’s about making technology accessible and purposeful. “I always try to make it easier for people to understand technology, and so we can deploy it at scale, and always look at the business case; return on investment is also very important.”
Despite this, Brooks says we have to accept that when it comes to AI there are always some hard-to-solve cases that can take decades to solve. “Regardless of how an AI system is deployed, there is always a long queue of special cases that take decades to discover and fix. The irony is that all of these fixes are made by AI itself.”
Brooks said this is a misconception, mainly because Moore’s Lawthat there will always be exponential growth when it comes to technology – the idea that if ChatGPT 4 Is it good, imagine what ChatGPT 5, 6 and 7 will be like. He sees a flaw in this argument, that technology doesn’t always grow exponentially despite Moore’s Law.
He used the iPod as an example. For a few iterations, its storage size actually grew from 10 to 160GB. If it had kept going at that pace, he predicted we’d have an iPod with 160TB of storage by 2017, but of course that didn’t happen. The models being sold in 2017 actually came with either 256GB or 160GB because, as he pointed out, nobody really needed more than that.
Brooks admits that LLMs could help to some extent with home robots, where they can perform specific tasks, especially with an aging population and not having enough people to care for them. But he says that could also come with some unique challenges.
“People say, ‘Oh, big language models will make robots able to do things they can’t do.’ That’s not the problem. The problem of being able to do something is about control theory and all kinds of other hardcore mathematical optimization,” he said.
Brooks explains that this could eventually lead to robots with language interfaces useful for people in care situations. “It’s not useful to tell an individual robot in a warehouse to go out and get something for an order, but for elderly care homes it could be useful for people to say something to the robot,” he said.
mit-robotics-pioneer-rodney-brooks-thinks-people-are-vastly-overestimating-generative-ai-techcrunch
